In the presented study, 60 decedents with a mean postmortem interval (PMI) of 2.9 days (SD 2.4, median 2.1, range 0.4–14.6) and a mean age of 72.3 years (SD 16.1, median 77.0, range 29–100 years) underwent partial autopsy with heart dissection, and the thickness of the ventricular walls was measured as introduced above. The sex ratio was 29:31 (f:m). The average postmortem body weight was 76.5 kg (SD 25.2, median 68.9, range 38.5–144.2), and the average body length was 170.1 cm (SD 12.0, median 169.5, range 140.0–196.0). The average BMI was 26.1 kg/m2 (SD 6.7, median 24.3 kg/m2, range 14.6–43.3). The interval between the echocardiographic examination and death or the time of dissection was on average 31.8 or 34.4 days respectively (SD 39.4 resp. 39.1, median 12.5 resp. 15.0, range 0.0–165.0 resp. 3.0–168.0). The average heart weight after dissection, rinsing, and drying was 488.6 g (SD 167.2, median 442.5, range 210–1016). In eight out of 60 decedents, sepsis was present at the time of death due to pneumonia or immunosuppression in leukemia according to the clinical records. The quality of echocardiography was reported as “very good” in 11 cases, as “good” and “adequate” in 18 cases, and as “satisfactory” in 13 cases (average 2.6, SD 1.1, median 3.0).

The echocardiographic and autopsy measurements were compared using the paired t-test. In addition, Pearson’s correlation coefficient and the intraclass correlation (ICC) were calculated. The results are shown in detail in Table 2. The Bland–Altman plots do not show a high level of agreement between the echocardiographic and the autoptic measurements (Fig. 3a and b). Deviations up to 10 mm could be detected when measuring the left ventricle. The largest differences were found for the measurements of the left ventricle using both dissection methods with the ventricle wall thickness being higher in the postmortem measurement.

a Bland–Altman plots illustrating the differences between the measurements against their mean values for the inflow-outflow method for all measured localizations. LVPW PLAX, left ventricular posterior wall thickness in parasternal long axis; LVPW SAX, left ventricular posterior wall thickness in short axis; RV subcostal, right ventricular wall thickness subcostal. b Bland–Altman plots illustrating the differences between the measurements against their mean values for the short-axis method are shown in the figure for all measured localizations. RV subcostal, right ventricular wall thickness subcostal; IVS PLAX, interventricular septum thickness in parasternal long axis; IVS SAX, interventricular septum thickness in short axis

The largest measurement differences were found for the measurements below the aortic valve. A measurement of 1 cm and 2 cm below the aortic valve resulted in a difference of 4.4 mm (95% CI [3.3; 5.5]) and 5.2 mm (95% CI [4.1; 6.4]) mm respectively between the echocardiographic (parasternal long axis) and the autoptic measurement. The differences for the measurements below the mitral valve are somewhat smaller—in particular, the measurement 1 cm below the mitral valve showed differences between 3.3 mm (95% CI [2.3; 4.4]) and 3.5 mm (95% CI [2.6; 4.5]) mm, depending on the comparison with the echocardiographic measurement plane. This measurement localization resulted in a positive Pearson correlation with a medium effect size (0.48) and a poor ICC of 0.22. For the right ventricle, slightly lower values were measured for the measurement 1 cm below the tricuspid valve than for the echocardiographic examination (− 0.2 (95% CI [− 1.3; 1.0])), whereas the opposite was true for a measurement 2 cm below the tricuspid valve (0.5 (95% CI [− 1.1; 2.0])). Overall, the comparison of the measurements of the right ventricle showed smaller differences, with small negative correlation coefficients and poor ICC showing poor agreement. In only three out of eight decedents clinically diagnosed with sepsis, measurement differences of up to 11 mm for the left ventricle and septum, and up to 5 mm for the right ventricle were observed.

While the measurement differences after using the short-axis dissection were 3.5 mm (95% CI [2.1; 4.9]) for the left ventricle, the measurements of the septum and the right ventricle showed smaller differences. In addition, Pearson’s correlations for the measurement in the short-axis dissection showed a medium effect size for the left and right ventricle (0.34 and 0.33 respectively), a large effect size for the septum (0.84), and a good agreement (ICC 0.78).

When considering the overall cohort, there was a tendency towards lower measurement differences with increasing PMI (> 2 days) with a small positive correlation for the measurements 1 and 2 cm below the aortic valve (0.25; 0.22) and 1 and 2 cm below the tricuspid valve (0.28; 0.29), and a small negative correlation for the measurements of the left (− 0.13) and right ventricle (− 0.19), and the septum (− 0.31) after using the short-axis dissection.

A further aim of the study was to determine whether postmortem measurements were able to confirm antemortem hypertrophy. The results are presented in Table 3. The diagnostic performance of all postmortem inflow-outflow measurements can be classified as a failure, as the area under the ROC curve is below 0.6 (an area under the ROC curve of 0.5 is equivalent to chance, while 1 indicates perfect discrimination [26]). This is also reflected in a sensitivity between 67 and 76% and a small specificity between 0 and 33%.

While measurements with the short-axis method for the left ventricle yielded similar results (sensitivity 75%, specificity 35%), the area under the ROC curve for the right ventricle (0.720) and septum (0.823) shows a fair or considerable diagnostic performance, respectively. Using the threshold for postmortem measurements presented in the literature, for 80% of patients with a hypertrophy diagnosed antemortem in the right ventricle this can also be confirmed postmortem after short-axis dissection (sensitivity 80%), leaving 20% of the patients with a hypertrophy antemortem undiagnosed. Sixty percent of patients without hypertrophy during their lifetime are recognized accordingly postmortem (specificity 60%). The probability of having had hypertrophy during lifetime if diagnosed postmortem using the published thresholds is 91% for the right ventricle, but the probability of not having had hypertrophy if diagnosed as such postmortem is 38%. The test diagnostics when using the septum measurements show opposite directions, with a test being more specific and less sensitive. Seventy-four percent of patients without hypertrophy antemortem are recognized accordingly postmortem, and 57% of patients with hypertrophy during lifetime could be confirmed postmortem after using short-axis dissection. The probability that persons did not have antemortem hypertrophy, when none was detected during autopsy was 85% for the septum and 90% for the left ventricle, while the probability of having had hypertrophy in the patient’s lifetime if diagnosed postmortem is only 15% for the left ventricle and 40% for the septum. Optimal cut-off values calculated in the ROC analysis resulted in postmortem threshold values with higher specificities and PPV (Table 3). Postmortem threshold values between 20 and 23 mm were found for the left ventricle and septum (specificity 89–100%), and between 7 and 8 mm for the right ventricle using the inflow-outflow method (specificity 80%). Using the short-axis method, the threshold given in the literature for the right ventricle could be confirmed.